An example of how we’re using Event Hubs and agents is that with some of our clients we don’t have direct access to some of their network infrastructure (such as to manage their firewall and collect configuration backups) so, rather than poking holes to give us access publicly or build a site-to-site tunnel between our edge and their edge, we have an Event Hub agent installed on their management server, and we can then send commands to it through the PSU server to have it make some calls to the firewalls’ XML APIs to gather the configuration output. This command is contained in a variable on the PSU side so that the result of the API query from the agent will be received on the PSU side for further processing.
As the way this is designed is fairly generic, we could do any number of things through the agent, such as sending it commands, dumping a script into a variable to send over to the agent for it to run locally (assuming any required PowerShell modules have been installed already that the script(s) will need to use), trigger a script that’s already on the management server, etc.